From EMBER to FIRE: predicting high resolution baryon fields from dark matter simulations with deep learning
Year: 2021
Type: article
Abstract: ABSTRACT Hydrodynamic simulations provide a powerful, but computationally expensive, approach to study the interplay of dark matter and baryons in cosmological structure formation. Here, we introduce the EMulating Baryonic EnRichment (EMBER) Deep Learning framework to predict baryon fields based on dark matter-only simulations thereby reducing computational cost. EMBER comprises two network architectures, U-Net and Wasserstein Generative Adversarial Networks (WGANs), to predict 2D gas and H i densities from dark matter fields. We design the conditional WGANs as stochastic emulators, such that multiple target fields can be sampled from the same dark matter input. For training we combine cosmological volume and zoom-in hydrodynamical simulations from the Feedback in Realistic Environments (FIRE) project to represent a large range of scales. Our fiducial WGAN model reproduces the gas and H i power spectra within 10 per cent accuracy down to ∼10 kpc scales. Furthermore, we investigate the capability of EMBER to predict high resolution baryon fields from low resolution dark matter inputs through upsampling techniques. As a practical application, we use this methodology to emulate high-resolution H i maps for a dark matter simulation of a $L=100\, \text{Mpc}\, h^{ -1}$ comoving cosmological box. The gas content of dark matter haloes and the H i column density distributions predicted by EMBER agree well with results of large volume cosmological simulations and abundance matching models. Our method provides a computationally efficient, stochastic emulator for augmenting dark matter only simulations with physically consistent maps of baryon fields.
Language: English
Subfield: Astronomy and Astrophysics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
Funders: National Science Foundation, National Aeronautics and Space Administration, Simons Foundation, Space Telescope Science Institute, Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung, Barcelona Supercomputing Center, Universität Zürich, Centro Svizzero di Calcolo Scientifico
Awards: G1419135201, G1687310753, Computation for the Endless Frontier, G2474591846, The Interscale Galactic Nuclei Simulations (IGNIS): Hyper-refined black hole growth and feedback in cosmological environments, G2731017927, CAREER: The Faint Frontier of Galaxy Formation, Aiming for the Parsec Scale - Star Formation and Feedback Processes in High Redshift Galaxies, AS THE FAINTEST LOWEST-LUMINOSITY AND MOST DARK-MATTER-DOMINATED EXTREMES OF THE GALAXY POPULATION DWARF GALAXIES PRESENT UNIQUE OPPORTUNITIES FOR STUDYING GALAXY FORMATION AND THE PROPERTIES OF DARK. ACCORDINGLY THEY HAVE BEEN OBSERVED IN DETAIL FROM THE GROUND AND (BY NASA MISSIONS) FROM SPACE WITH THE HOPES OF UNRAVELING HOW DWARF GALAXIES FORM THE EFFECTS OF REIONIZATION ON GALAXY FORMATION AND WHETHER SIGNATURES OF THE PARTICLE NATURE OF DARK MATTER (DM) INTERACTIONS CAN BE OBSERVED. SUCH WORK HAS GAINED INIMPORTANCE AS EFFORTS TO DIRECTLY DETECT DM HAVE SO FAR YIELDED ONLY UPPER LIMITS TO THE INTERACTION BETWEEN DM AND NORMAL MATTER LEAVING ASTROPHYSICAL TESTS AS THE PRIMARY MEANS OF INVESTIGATING THE NATURE OF DM. WE PROPOSE TO UNDERTAKE AN EXTENSIVE YET FOCUSED PROGRAM OF COSMOLOGICAL HYDRODYNAMIC SIMULATIONS AIMED AT UNDERSTANDING THEFORMATION OF DWARF GALAXIES. WE WILL FOCUS ON THE INTERPLAY BETWEEN GALAXY FORMATION AND DARK MATTER IN THESE GALAXIES POINTING TOWARD SPECIFIC OBSERVABLES TO DISENTANGLE THE EFFECTS OF GALAXY FORMATION PHYSICS FROM THE EFFECTS OF DM PHYSICS. OUR SIMULATION SUITE WILL EXPLORE COLLISIONLESS COLD DARK MATTER (CDM) AND BROAD CLASSES OF ALTERNATIVES WHERE DM HAS A NON-NEGLIGIBLE FREE-STREAMING LENGTH AND / OR SELF-SCATTERING CROSS SECTION. THE NOVEL ASPECTS OF THE PROPOSED WORK WILL INCLUDE: (1) A MODERN TREATMENT OF BARYONIC PHYSICS USING GIZMO A NEW CODE THAT USES ACCURATE MESHLESS METHODS FOR HYDRODYNAMICS (2) FEEDBACK IN REALISTIC ENVIRONMENTS (FIRE) A SUITE OF GALAXY FORMATION PARAMETRIZATIONS WITH WELL-TESTED EXPLICIT IMPLEMENTATIONS OF STELLAR FEEDBACK (3) AN EXPLORATION OF REALISTIC MODELS OF DM BEYOND CDM BASED ON AN EFFECTIVE THEORY OF STRUCTURE FORMATION WITH FULL BARYONIC PHYSICS AND (4) DETAILEDMOCK OBSERVATIONS OF THE SIMULATIONS IN ORDER TO IDENTIFY SPECIFIC DISTINGUISHING TESTS FOR CDM AND ITS ALTERNATIVES.OUR RESEARCH WILL PROVIDE A FRAMEWORK WITHIN WHICH ASTROPHYSICAL INFERENCES ABOUT THE NATURE OF DM CAN BE FAIRLY AND SELFCONSISTENTLY TESTED. CRUCIALLY WE WILL EXPLORE UNCERTAIN ASPECTS OF GALAXY FORMATION AND DM PHYSICS IN A CONTROLLED MANNER FOCUSING ON THE HALO MASS RANGE -- 9.5<LOG10(M/M_SUN)<10.5 -- WHERE CURRENT MODELS ARE MOST UNCERTAIN. THIS WILL MARK A MAJOR EFFORT TOSYSTEMATICALLY STUDY DIFFERENT DM MODELS COMBINED WITH REALISTIC TREATMENTS OF GALAXY FORMATION PHYSICS THAT ARE IMPLEMENTED IN AN IDENTICAL WAY FOR THE VARIOUS DM MODELS. THE RESULTS OF THE PROPOSED RESEARCH WILL BE DRAMATIC IMPROVEMENTS IN THE UNDERSTANDING OF HOW BARYONIC VERSUS DM PHYSICS AFFECT ASTROPHYSICAL OBSERVABLES. SUCH A STUDY IS VERY TIMELY AS OBSERVATIONS OF DWARF GALAXIES IN ANDBEYOND THE LOCAL GROUP HAVE PROGRESSED SUBSTANTIALLY IN RECENT YEARS WITH NEW REVELATIONS ABOUT DWARFS' DM CONTENT STAR FORMATION HISTORIES AND STELLAR METALLICITIES CHALLENGING PREVIOUS INTERPRETATIONS., Collaborative Research: Starless Dark Matter Halos as a Definitive Test of Dark Matter Models, G58316358, G5941523914, G6025057615, G7737588766, Aiming for the Parsec Scale - The Rise of Massive Galaxies and their Supermassive Black holes, G890379361
OpenAlex ID: W3208715870